Data Mesh is the latest new buzzword in data architecture - what is it?

In a large organisation there is usually lots of data from many sources. This may be useful to the organisation, but often it is hard to get access to and understand. Making data available to everybody who needs it in a usable format is hard, and companies often spend money and time trying to do this with various centralised data warehouse projects. These often fail.

Data Mesh is a way of making data available to others by treating it as a product and managing it in a decentralised way. The term was first coined by Zhamak Dehghani who works at the consultancy Thoughtworks. The ideas have also been explained by Chris Riccomini.

The ideas behind the approach are borrowed from how modern software is developed, as networks of loosely cooled micro-services managed in a decentralised way. Each service is an API that provides access to a set of data that is well documented and managed. Sounds good, but will it work in practice?